A former Anthropic researcher has warned that the rapid development of artificial intelligence could create serious risks for humanity.
Jacob Coxon, 27, said people working inside the AI industry are deeply concerned about how quickly the technology is advancing.
Coxon spoke to the BBC after his resignation post about AI safety attracted widespread attention online.
‘People are genuinely frightened’
Coxon said the current pace of AI development is moving faster than the ability of governments and companies to manage its risks.
He believes the consequences could become catastrophic if the industry does not slow down.
“I believe that if we don’t slow down at the current rate of progress, there is a strong chance that we could all die in the immediate future.”
The researcher previously worked at OpenAI before joining Anthropic.
He said many AI workers are serious about calls for regulation because they feel trapped in a race to build increasingly powerful systems.
Call for international cooperation
Coxon’s warning comes as Anthropic CEO Dario Amodei pushes for a more cautious approach to AI development.
Amodei recently argued that the technology itself should continue to develop. However, he said its risks require urgent attention.
He called for companies and governments to have enough time to establish effective safeguards.
Coxon supports the idea of slowing AI development.
However, he said the process must involve China and other major technology powers.
Without international coordination, he warned, countries could enter a dangerous race to develop more advanced AI systems.
What could an AI disaster look like?
Coxon said it is difficult to predict exactly how an AI catastrophe could unfold.
One possibility involves large numbers of AI bots working together with enormous computing power.
Such systems could potentially gain control over parts of the internet, according to the researcher.
Coxon said he believes this type of scenario could become realistic within six months to one year.
His prediction remains controversial, with other technology leaders disputing some of the industry’s most extreme warnings.
Anthropic defends its safety measures
Anthropic has defended its approach to AI safety following Coxon’s departure.
The company said it recognises that artificial intelligence could bring major benefits while also creating unprecedented risks.
It said it continues to develop safeguards and conduct extensive tests on its models.
Anthropic also said it publishes safety research to allow its work to be examined by others.
The company supports cooperation across the AI industry to establish clear and verifiable rules for releasing powerful models.
Researchers divided over AI risks
Coxon is not alone in warning about the possible dangers of advanced AI.
Anthropic scientist Evan Hubinger has also expressed concerns about the possibility of AI becoming an existential threat.
Hubinger said he personally believes there is more than a 10% chance that AI could eventually kill all humans within the next decade.
Meanwhile, computer scientist Geoffrey Hinton, widely known as the “Godfather of AI”, has said a 10% risk of AI causing human extinction is not unreasonable.
AI’s potential benefits
Despite his concerns, Coxon said he remains hopeful about the technology.
He said AI researchers genuinely want to see the positive side of artificial intelligence.
Potential benefits include helping to solve diseases and improving people’s quality of life.
For Coxon, the challenge is finding a way to develop those benefits without allowing the technology to advance beyond effective human control.
The growing debate has placed renewed attention on AI regulation, international cooperation and the need for stronger safety measures.
BBC
